Concerns that treatment for heroin dependence using naltrexone may increase suicide rates during treatment and fatal overdoses posttreatment have been expressed. There is also disquiet about mortality during induction onto methadone. We assessed mortality during specific periods following treatment with naltrexone implants or methadone. Data were assembled using the Western Australian Data Linkage System. The methadone cohort comprised all those who started methadone in Western Australia during 2001-2002: The naltrexone cohort comprised all Western Australian heroin-dependent persons who received their first implant in 2001-2002. There were 15 (2.7%) deaths in the methadone cohort (n = 553) and 6 (1.8%) deaths in the naltrexone cohort (n = 341). Mortality rates for the "initial 14-day period," "stable treatment," and "overall" were 94.47, 0.0, and 5.83 deaths/1,000 person-years for the methadone group. In the naltrexone group, the rates "during first treatment (0-6 months)," "post first treatment," and overall were 0.0, 4.21, and 3.76 deaths/1,000 person-years. The age-standardized mortality rate ratio for naltrexone compared to methadone was 0.645 (95% confidence interval = 0.123-1.17). Increased mortality during induction onto methadone was confirmed. Evidence relating naltrexone to either increased suicide or overdose was not found. Overall mortality rates for naltrexone implant were similar to those for methadone, but increased mortality during methadone induction was avoided.